**** BEGIN LOGGING AT Fri Jul 06 02:59:58 2012 Jul 06 07:08:02 can anyone help me how to compile Xfbdev in OE? Jul 06 09:01:03 morning all Jul 06 12:56:36 hi bluelightning, all Jul 06 13:01:06 hi pb_ Jul 06 13:01:13 quiet in here today Jul 06 13:10:25 pb_: hey there Jul 06 13:29:43 hi ant_work Jul 06 13:29:52 bluelightning: yeah, indeed Jul 06 13:30:03 are you underwater down there? Jul 06 13:30:07 * ant_work is ready to fire questions Jul 06 13:30:35 pb_: your C karma is still very high, isn't? Jul 06 13:30:46 well, so-so Jul 06 13:30:53 what is your question? Jul 06 13:31:21 I've almost solved, need a confirmation about array of char pointers Jul 06 13:32:14 see the first function https://github.com/andrea-adami/kexecboot/commit/af94b1b538fcd30335bd05f833ecd156e0d666c0 Jul 06 13:32:51 Khem did help me to solve first big doubts, I'm still dubious about arg count Jul 06 13:33:30 exactly about the necessity of a null pointer at the end of the array Jul 06 13:34:30 pb_: no, no flooding in the southeast that I'm aware of... Jul 06 13:36:40 ant_work: I'm not familiar with fexecw(). Is that a standard function? Jul 06 13:37:00 no, is fork-execv-wait Jul 06 13:37:30 the poit is, the getopt should be done by the called 'ubiattach' function Jul 06 13:38:25 this fexecw was already used in the code , for kexec. I've reused it for ubiattach. Jul 06 13:39:12 well, your ubiattach_argv[1] looks a bit dubious Jul 06 13:39:27 and [2] for that matter. Jul 06 13:39:29 yes..is a 'string' Jul 06 13:39:52 yeah, I wouldn't expect that to work Jul 06 13:40:13 heh...that's why I reserved more arg[] in declaration Jul 06 13:40:26 ubiattach -m 3 -O 512 Jul 06 13:40:29 heh Jul 06 13:40:36 argc = 5 Jul 06 13:40:50 still, happens to work, Jul 06 13:40:57 fair enough Jul 06 13:40:59 I imagine because I'm passing pointers to literal Jul 06 13:41:15 but I smell bad code here :p Jul 06 13:41:36 so, anyway, what exactly was the question? Jul 06 13:41:48 would you pass 2 args or 5 ? Jul 06 13:41:52 5 Jul 06 13:41:56 I think I know now... Jul 06 13:41:58 :) Jul 06 13:42:16 { "ubiattach", "-m", "3", "-O", "512", NULL } Jul 06 13:42:22 corollary..is a 6th null necessary in the initialization? Jul 06 13:42:29 =) Jul 06 13:42:30 you do need a NULL at the end, yes Jul 06 13:42:58 so getopt will stop parsing on first NULL, isn't? Jul 06 13:42:58 since you don't pass argc, and there is no other way for the called function to know when to stop reading Jul 06 13:43:05 I see Jul 06 13:43:57 well, what actually happens (usually) is that the C runtime code in the exec'd program counts the arguments and sets argc appropriately when it calls main(), then getopt uses whatever value of argc it received. Jul 06 13:44:07 but the NULL is still there and getopt could indeed use that as well if it wanted Jul 06 13:44:14 ok, ubiattach.c does getopt_long Jul 06 13:45:32 in the first vesrions of the code, I had some null in-between and a last "/dev/ubi_ctrl" tag. This was because klibc's getopt needs this syntax (posix conform ?) Jul 06 13:45:50 now, I got rid of this last arg, being is default in the ubiattach code Jul 06 13:46:04 so I can have a variable number of args... I hope Jul 06 13:46:08 ubiattach -m3 Jul 06 13:46:19 ubiattach -m 3 -O 512 Jul 06 13:46:29 should both work *after* your correction Jul 06 13:48:05 right Jul 06 13:48:12 many many thx! Jul 06 13:49:21 no problem Jul 06 13:57:14 pb_: googling for a solution I got overconfused by http://blog.flameeyes.eu/2007/12/array-of-pointers-and-array-of-arrays Jul 06 13:57:28 that's aread for a few of you ;) Jul 06 15:24:11 otavio: ping Jul 06 15:49:15 hi all Jul 06 15:49:35 where is lauched x in oe/ngstrom? Jul 06 15:49:41 where is lauched x in oe/Angstrom? Jul 06 16:36:20 hey, prize question: i copied my whole oe toolchain from ubuntu 10.10 to 12.10 .. it looks working and builds as exspected , yet the ipk md5sum are different .. is that exspected ? Jul 06 16:36:52 hm Jul 06 16:37:40 i wonder i ipkg packaging will bring in some strings from the real host and thats the difference ? Jul 06 16:38:40 but .. generally speaking.. if the chain builds and packages .. not all much be totally wrong right ? Jul 06 16:39:44 it shouldnt because host stuff is not used normaly Jul 06 16:40:04 but I dont know excactly what opkg is doing at package creation Jul 06 16:47:10 ah hang on .. i used packages which inbuild a date stamp at compile time .. my foo Jul 06 16:47:19 looking for a different pacakge Jul 06 20:56:02 d.setVar("PACKAGES", "%s kernel-devicetree" % packages) Jul 06 20:56:32 I feel like this seets the PACKAGES variable to contain kernel-devicetree more than once ,,, Jul 06 21:54:57 Crofton: you could use appendVar there Jul 06 21:56:20 d.appendVar("PACKAGES", "kernel-devicetree") **** ENDING LOGGING AT Sat Jul 07 02:59:58 2012